Germonpre Et Al. Palaeolithic Dog-libre

8/12/2019 Germonpre Et Al. Palaeolithic Dog-libre

1/18

Fossil dogs and wolves from Palaeolithic sites in Belgium, the Ukraineand Russia: osteometry, ancient DNA and stable isotopes

Mietje Germonpre a,*,1, Mikhail V.Sablin b,1, Rhiannon E. Stevens c, Robert E.M. Hedges d,Michael Hofreiter e, Mathias Stiller e, Viviane R. Despres e,2

a Department of Palaeontology, Royal Belgian Institute of Natural Sciences, Vautierstraat 29, 1000 Brussels, BelgiumbZoological Institute RAS, Universitetskaya nab. 1, 199034 Saint-Petersburg, Russiac McDonald Institute for Archaeological Research, University of Cambridge, Downing Street Cambridge CB2 3ER, UKd Research Laboratory for Archaeology and the History of Art, University of Oxford, Dyson Perrins Building, South Parks Road, Oxford, OX1 3QY, UKe Max Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, D-04103 Leipzig, Germany

a r t i c l e i n f o

Article history:

Received 6 May 2008

Received in revised form

23 September 2008

Accepted 26 September 2008

Keywords:

Upper Palaeolithic

Canidae

Dog

Skull

Ancient DNA

Stable isotopes

a b s t r a c t

Using multivariate techniques, several skulls of fossil large canids from sites in Belgium, Ukraine andRussia were examined to look for possible evidence of the presence of Palaeolithic dogs. Reference

groups constituted of prehistoric dogs, and recent wolves and dogs. The fossil large canid from Goyet(Belgium), dated at c. 31,700 BP is clearly different from the recent wolves, resembling most closely theprehistoric dogs. Thus it is identified as a Palaeolithic dog, suggesting that dog domestication had already

started during the Aurignacian. The Epigravettian Mezin 5490 (Ukraine) and Mezhirich (Ukraine) skulls

are also identified as being Palaeolithic dogs. Selected Belgian specimens were analyzed for mtDNA andstable isotopes. All fossil samples yielded unique DNA sequences, indicating that the ancient Belgianlarge canids carried a substantial amount of genetic diversity. Furthermore, there is little evidence for

phylogeographic structure in the Pleistocene large canids, as they do not form a homogenous genetic

group. Although considerable variation occurs in the fossil canid isotope signatures between sites, the

Belgian fossil large canids preyed in general on horse and large bovids.2008 Elsevier Ltd. All rights reserved.

1. Introduction

The evolutionary origin of the dog from wolves is well estab-lished via morphological (Benecke, 1987; Clutton-Brock, 1997;Morey,1992; Nobis, 1986; Olsen,1985) and genetic data (Savolainen

et al., 2002; Vilaet al., 1997). Between 14,000 and 10,000 years agodogs are known from Western Europe (Nobis, 1986; Chaix, 2000),Southern Europe (Altuna et al., 1985; Vigne, 2005), the Near East(Davis and Valla, 1978; Tchernov and Valla, 1997), the Russian Plain

(Sablin and Khlopachev, 2002, 2003) and Kamchatka (Dikov, 1996).Dogs accompanied humans into the New World 12,00014,000years ago (Fiedel, 2005; Leonard et al., 2002). At that time theancestral population of dogs in Eurasia was probably already large

(Leonard et al., 2002). According toSavolainen et al. (2002)mostrecent dog populations have a common origin from a single genepool in East Asia, descending from approximately five mtDNA

lineages. Genetic results suggest a much older origin of dogs thanindicated by prehistoric finds (Vilaet al.,1997). mtDNAdata indicate

that the domestication of the dog started either at around 40,000years ago or at around 15,000 years ago (Savolainen et al., 2002).According toLindblad-Toh et al. (2005), an ancient genetic bottle-

neck accompanying the domestication of dogs occurred around27,000 years ago. Withthis in mind, thelow frequency of recogniseddog skulls in Upper Palaeolithic sites is somewhat surprising. In ouropinion, it is likelythat a numberof Palaeolithic dog remainshaveso

far not been recognized. We conducted an osteometric analysis offossil large canids, which possibly could be eitherdog or wolf, foundin Belgian, Ukrainian and Russian sites with the aims of identifyingand distinguishing Palaeolithic dogs from fossil wolves. Our

hypotheses are that changes in dog morphology compared to wolfmorphology appeared rather abruptly, that they were linked to theeffects of domestication and that these changes became fixed in thedog population. If evidence cannot be not found to support these

hypotheses, the alternative hypothesis would then be thatsubstantial morphological differences were present betweenPleistocene wolf populations, before domestication, and betweenlineages of wolves thatled later onto recentwolves and dogs. Inthis

situation we would expect to see a gradual morphological change

* Corresponding author. Tel.: 32 2627 44 64.

E-mail address: [email protected](M. Germonpre).1 These authors contributed equally to this work.2 Present address: Max Planck Institute for Chemistry, Department of Biogeo-

chemistry, D-55128 Mainz, Germany.

Contents lists available atScienceDirect

Journal of Archaeological Science

j o u r n a l h o m e p a g e : h t t p : / / w w w . e l s e v i e r . c o m / l o c a t e / j a s

0305-4403/$ see front matter 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.jas.2008.09.033

Journal of Archaeological Science 36 (20 09) 473490
mailto:[email protected]://www.sciencedirect.com/science/journal/03054403http://www.elsevier.com/locate/jashttp://www.elsevier.com/locate/jashttp://www.sciencedirect.com/science/journal/03054403mailto:[email protected]


2/18

from wolves to dogs. Theosteometrics of thefossil large canids werecompared through univariate and multivariate analyses to those of

prehistoric dogs, recentwolves and recent dogs in order to establishto which group the fossil specimens belong. We did not distinguisha priori a group of fossil wolves. A similar approach was developed

by Morey (1986) on Amerindian dogs and Benecke (1987) onEuropean Palaeolithic finds. We studied some of the same speci-mens as the latterinvestigation, but added material unknown at thetime of Beneckes analysis.

Genetic analysis was undertaken on Belgian fossil large canidswith the goal to compare the analyzed Belgian specimens with therecent dog and wolf mtDNA haplotypes described to date.

Isotopic analyses were conducted with the aims of recon-structing the diet of the Belgian fossil large canids and comparingthem to those of other Pleistocene fossil canids. This dietaryinformation may provide some insight into the relationship

between humans and the Pleistocene canids. Whether identified asPalaeolithic dogs or fossil wolves by the morphometric and DNAanalyses, we aimed to consider the diets of the fossil canids incontext of the speculated broadening of human diets during the

mid/late Upper Palaeolithic to include freshwater resources(Richards et al., 2001).

2. Material and methods

2.1. Osteometric analysis

Our sample comprised of 117 skulls of recent and fossil largecanids (Table 1). The AMS age or the supposed age of the skulls isgiven inTable 1. The Belgian canids originate from six Palaeolithiccaves and one postglacial cave, located in the Condroz, a region in

the south of Belgium (Fig. 1).The Goyet cave is situated in a limestone cliff in the Samson

valley, a tributary of the Meuse River. The cave consists of severalchambers in which a large number of Middle and Upper Paleolithic

artefacts were discovered along with numerous remains of ice age

mammals (Dupont, 1873). Manyof the fossil bones are broken, havecut marks, or display traces of ochre (Germonpre, 1996; Germonpreand Hamalainen, 2007). The Palaeolithic artefacts date from the

Mousterian, Aurignacian, Gravettian, and Magdalenian, whichindicates recurrent occupations of the cave from the Pleniglacialuntil the Late Glacial. Unfortunately, it is not always clear fromwhich horizon the artefacts and bones originated (Dewez, 1987;

Otte and Groenen, 2001; Ulrix-Closset, 1975). Aurignacian ivorybeads were discovered in Horizon 3 (Otte, 1979). This horizon isa palimpsest of multiple occupations (Miller, 2001). Other spec-tacular finds include batons de commandement, needles, perfo-

rated teeth, a bone harpoon and shell necklaces from theMagdalenian (Horizons 1 and 2) (Dewez, 1987). Deeper inside thecaveDupont (1873)distinguished a fourth and fifth horizon con-

taining mainly bones from cave bear and cave lion. The fossil canidskull found during Edouard Duponts excavations in the 1860s hasan AMS age of c. 31,700 BP. According to Duponts unpublishednotes, the skull was found in a side gallery of the cave, in Horizon 4,

together with remains from mammoth, lynx, red deer and largecanids.

The Trou des Nutons (Furfooz), Trou Bailleux and Trou de laNaulette caves are situated in limestone cliffs on the banks of the

river Lesse, a tributary of the Meuse. In the 1860s, Dupont exca-vated in Trou des Nutons cave a partly associated skeleton of a largecanid that he identified as wolf (Dupont, unpublished notes). Theright humerus displays cut marks; the skull has an AMS age of

21,800 BP. However, the main bone horizon produced Magdalenianartefacts and a cut-marked phalanxof horse has been dated by AMS

to 12,630 y BP (Charles, 1998). Trou de la Naulette is a famousNeanderthal site excavated by Dupont in the 1860s (Dupont, 1873).

According to the notes of Dupont, the fossil canid skullwas found inthe Second Horizon, the same containing the Neanderthal remains.Trou Baileux (Balleux) was excavated in 1866 and in the 1980s

(Dupont,unpublished notes;Depaepe, 1988). Dupont (unpublishednotes) discovered remains from beaver, red deer, roe deer, horse,bison, sheep/goat and pig. The species present point to a postglacial

age forthis assemblage. The canid skullmost probably forms part ofthe postglacial assemblage discovered by Dupont; its appearance issimilar to that of the bones from this assemblage.

The cave of Grands Malades was situated on the left bank of theriver Meuse. Several bones of large canids were discovered at thesite, as were Mousterian artefacts (Ulrix-Closset, 1975).

The Ukrainian and Russian fossil large canids are from the

Russian Plain and Siberia (Fig. 1). One skull from Siberia was foundin the permafrost in fluvial deposits on the bank of a tributary of theAnabar River (Yakutia). This isolated find is not related to any

prehistoric site and therefore it is assumed that this specimen isfrom a fossil wolf. The other skulls were discovered at UpperPalaeolithic sites from the Russian Plain. The fauna at the Gravettiansite of Avdeevo includes mammoth, rhinoceros, horse and reindeer.

The large quantity of arctic fox and wolf bones suggests the exis-tence of fur hunting. Most of the artefacts and art pieces are made

from mammoth tusks (Gvozdover, 1995). The Epigravettian Mezinis well known for its round mammoth bone dwelling. At Mezherich,

also dating from the Epigravettian, four mammoth bone dwellingsare present (Pidoplichko, 1998; Soffer, 1985).

Dogs and wolves were used as reference groups (Table 1). The

first reference group consisted of European prehistoric dogs, con-taining the two Palaeolithic dogs from the Epigravettian Eliseevich Isite (Russian Plain), with an age of around 13,900 BP (Fig. 1). Here,remains of at least eight mammoth bone complexes and large

quantities of worked ivory were discovered (Sablin and Khlo-pachev, 2002, 2003). The most complete dog skull (447) was foundin a hearth deposit, near a concentration of mammoth skulls(Polikarpovich, 1968). Its braincase has been perforated on the left

and right side (Sablin and Khlopachev, 2002, 2003). Cut marks are

present on the zygomatic and frontal bones. With exception of thecanines and some premolars, all its teeth are missing. In additionthe left and right carnassials were apparently removed by

damaging the alveoli.In order to have a larger reference group, that contained more

than two specimens, we added to the prehistoric dog group threeyounger and smaller dogs: the Epipalaeolithic dog of Saint-Thibaud,

France (Chaix, 2000) and the Mesolithic dogs fromthe German sitesBedburg (Street, 1989) and Senckenberg (Degerbl, 1961).

The second reference group was made up of recent dogs. Breedsof dogs whose genetic relationships are known from molecular

marker studies were specifically selected (Parker et al., 2004).Phylogenetic analysis separated several dog breeds with ancientorigins (Chow Chow, Siberian Husky) from a larger group of breeds

with modern origins (Parker et al., 2004). Thus the recent dogs inour study were divided into two sub-groups: the recent archaicdogs (including Chow chow and Siberian Husky) and recent otherdogs. The other modern breeds appear to represent a more recentradiation from shared European stock (Parker et al., 2004). We

selected large animals with a wolf-size skull (Irish Wolfhound,Mastiff, Tibetan Mastiff, Great Dane, Doberman Pinscher, GermanShepherd Dog) along with Malinois and Rottweiler, comparable in

size to the Husky. We also added one skull of a Central AsianShepherd Dog as an unclassified specimen.

Recent or historical wolves from Belgium, northwestern Russia,Caucasus, Jamal, Yakutia, Kamchatka and the Far East formed the

last reference group. Two skulls from wolves kept in captivity inBelgian zoos were considered as unclassified specimen.

Only animals with the permanent dentition in place andcomplete fusion of the dorsal cranial sutures were considered in

M. Germonpreet al. / Journal of Archaeological Science 36 (2009) 473490474


3/18


4/18

check the position of the fossil large canids versus all referencegroups. To reduce the effect of size between the different largecanid groups the following indices were used: the carnassial

index (P4CL/TL) and the braincase width index (GWbrc/TL) asdefined byLawrence and Bossert (1967), the snout index (VL/TL)as described by Harcourt (1974), the tooth-row length index(ALP1-M2/TL) as defined by Hell and Paule (1982), the greatest

snout width index (GW pal/VL) and the minimal snout widthindex (MWpal/VL) similar to Harcourts (1974) snout widthindex. The use of indices in the study of shape variations hasbeen criticized, partly because they are not always independent

of size (Atchley et al., 1976). However, since others have defendedthe use of such ratios for providing a good description of shape

(Corrucini, 1987; Van Valkenburgh and Wayne, 1994), weconsider their use justified. The indices in our study were log 10-transformed.

A linear discriminant function analysis (DFA) was carried out onlog-transformed ratios in order to assign the unclassified fossillarge canids to one of the reference groups.

2.2. Genetic analysis

In the phylogenetic analyses all ancient canid sequences werecompared to the wolf and dog sequences available in Genbank. A

median-joining network was constructed using the softwareNetwork version 4.5.0.0. byBandelt et al. (1999).Table 2lists the

Moscow

Kiev

Danube

Dniep

er

Desn

a

Dniester

9

0 100 200 300 km

Meuse

Lesse

Samson

Belgium

Germany

France 0 30 km

Luxemburg

Vesdre

Ourthe

BLACK SEA

32

1

54

6

7

8

10

Fig. 1. Map showing the most important sites discussed in the text. 1, Goyet; 2, Furfooz (Trou des Nutons and Trou du Frontal); 3, Trou de Chaleux; 4, Trou Balleux; 5, Trou de la

Naulette; 6, Grands Malades; 7, Eliseevich I; 8, Mezin; 9, Mezhirich; 10, Avdeevo; 11, Yakutia (Anabar basin).



5/18

Belgian bone material used for mtDNA-analysis with their AMS or

supposed age. Six specimens from Goyet cave and one from Troudes Nutons cave were analyzed. For these samples we amplified57 bp of the mitochondrial control region using the primers HVR1-wolf-F 50-ATA TTA TAT CCT TAC ATA GGA CAT-30 and HVR1-

wolf-R 50-ATT AAG CCC TTA TTG GAC T-30 spanning the nucleotidepositions 15,61215,668 of the dog mitochondrial genome (Gen-Bank accession number U96639). The chosen region comprised

a total of 30 polymorphic nucleotide positions and thereforecontains the majority of informative nucleotide positions in themitochondrial control region of wolves and dogs. Samples wereextracted following published protocols and guidelines (Hofreiter

et al., 2001; Paabo et al., 2004) and the DNA was amplified using5 ml of extract and 2 units of AmpliTaq Gold (Applied Biosystems) ina final reaction volume of 20 ml, consisting of 2.5 mM MgCl2, 1.25reaction buffer, 1 mM dNTPs and 0.25mM of each primer. Extraction

and amplification of ancient DNA was carried out in a laboratoryexclusively dedicated to ancient DNA work as described in Paaboet al. (2004). At least one mock extraction control and a minimum

of two PCR water controls were carried along during each extrac-tion and PCR. We performed at least two independent primaryPCRs from each sample, cloned the amplification products andsequenced a minimum of six clones per product in order tocorrectly assign each nucleotide possibly affected by ancient DNA

miscoding lesions. Except for the two samples Goyet 3 andGoyet 4, for which not enough bone material was available, allsequences were further replicated by sequencing at least anothersix clones of one or two additional amplification products derived

from a second independent DNA extraction. For further details seeStiller et al. (2006)also referring to the accession numbers (Goyet16: DQ852644DW852649 and Trou des Nutons 1: DQ852650).

Phylogenetic analyses on the mtDNA results were performed

using the programs MEGA3.1 (Kumar et al., 2004) and Network

(Bandelt et al., 1999).

2.3. Stable isotope analysis

Stable isotope analysis of bone collagen has been used exten-sively as a quantitative technique for reconstructing past animal

diets (Bocherens et al., 1996; Richards and Hedges, 1999; Schwarczand Schoeninger, 1991). These palaeodietary reconstructions arebased on the principle that food sources contain different isotopicsignatures, which are passed along the food chain to their

consumers.Bone collagen carbon (d13C) and nitrogen (d15N) isotope values

reflect the average isotope composition of the dietary proteinconsumed during the last few years of an animals life (Ambrose

and Norr, 1993; Tieszen and Fagre, 1993). Bone collagend

13

C valuesreveal the source of carbon, e.g. marine or terrestrial (Schoeninger

et al., 1983), or C3 versus C4 plants (Vogel and Van der Merwe,

1977). Marine fish (including anadromous fish such as salmon)have substantially higher d13C values than terrestrial herbivores,whereas freshwater fish can have more negative d13C values thanterrestrial and marine fauna. This is because marine plants (and

thus fauna) obtain most of their carbon from dissolved oceanbicarbonate (Hoefs, 1997), whereas in freshwater environmentscarbon can be from both geological sources and the atmosphere

(Fry and Sherr, 1984; Richards et al., 2001). Bone collagen d15Nvalues reflect the position of an animal in the food chain, as 35&enrichment in d15N occurs with each trophic level (Bocherens andDrucker, 2003; Peterson and Fry, 1987). Thus terrestrial herbivores

have lower d15N values than omnivores and carnivores from thesame ecosystem. Freshwater and marine food chains are longerthan terrestrial food chains; therefore the d15N values of aquaticorganisms (e.g. fish) are typically higher than those of terrestrial

herbivores (Katzenberg and Weber, 1999).Although diet is often the primary factor determining bone

collagen isotope values, small-scale isotopic variation may be

caused by other parameters such as climate and local environ-ment (Stevens and Hedges, 2004; Van Klinken et al., 1994; Vogel,1978).

Stable isotope analysis was performed on 10 Belgian fossil largecanids (Table 2). Five of these specimens came from the site Goyet

horizons 1 and 2. AMS dates on bones of various species from thesehorizons have almost exclusively produced Late-glacial ages (Ger-monpre, 1997; Dalen et al., 2007). Material from Trou de Chaleux,Trou du Frontal (Furfooz) and Trou des Nutons (Furfooz) was also

examined. Trou de Chaleux is located in a limestone cliff above theLesse River. Inside the cave, one major bone horizon containinga wealth of Magdalenian artefacts was discovered. The fauna isdominated by horse remains; other herbivores include reindeer

and musk ox (Charles, 1998; Dupont, 1873). Two cut-marked horse

bones and one musk ox bone excavated byDupont (1873)yieldedan AMS age of circa 12,850 B.P. (Charles, 1998; Hedges et al., 1993,1994). Besides remains of mammals, this cave also yielded bird and

fish remains (Dupont, 1873; Van Neer et al., 2007).Trou du Frontal (Furfooz) is located on the Lesse River near Trou

de Chaleux. This cave yielded a Magdalenian assemblage, including

remains of horse and reindeer. A cut-marked horse bone has anAMS age of 12,800 BP. Furthermore, also a significant proportion ofpostglacial material was excavated by Dupont in the 1860s. Alongwith remains of mammals, bird and fish remains were also

discovered at Trou du Frontal (Dupont, 1873; Van Neer et al., 2007).A selection of herbivores from the Late-glacial horizons at these

three sites were isotopically analyzed to allow assessment of theLate-glacial canid isotope results.

Two bones were samples from the associated skeleton of thePleniglacial wolf from Trou des Nutons (Table 2). Published isotope

Table 2

List of bones of the large canids used in this study for stable isotopes and mtDNA-analysis.

Site Age Isotopes Av. d13C Av. d15N Av. C/N %C %N mtDNA Material Remarques Reference

Belgium

Goyet A1 2812-10 13,680 60 BP G- 5 19.8 6.5 3.2 42.8 15.7 G-5 Bone KIA-25296 This study: DNA and isotopes

Goyet A1 2812-9 Late Glacial G-2 19.3 6.6 3.2 40.9 15.1 G-2 Bone This study: DNA and isotopes

Goyet A1 2812-8 Late Glacial G-7 19.6 8.9 3.2 42.3 15.2 Bone This study: isotopes

Goyet A2 2760-1 Late Glacial G-1 19.9 5.5 3.3 60.1 21.6 G-1 Bone This study: DNA and isotopes

Goyet A2 2760-20 Late Glacial G-8 19.4 8 3.2 37.3 13.7 Bone This study: isotopesGoyet A3 2240-1 Pleniglacial G-4 Bone This study: DNA

Goyet A3 2240-2 Pleniglacial G-3 Bone This study: DNA

Goyet B4 2860-2 24,780 140 BP G-6 Bone KIA-25297 This study: DNA

Trou du Frontal 2468-1 Postglacial? TF-1 21.3 9.3 3.3 42.7 15.3 Bone This study: isotopesTrou du Frontal 2468-2 Late Glacial? TF-2 18.9 9.2 3.3 40.4 14.4 Bone This study: isotopes

Trou des Nutons 2559-1 21,810 90 BP TN- 1 18.7 9.6 3.1 42.5 15.1 TN-1 Skull KIA-25298 This study: isotopes and DNA

Trou des Nutons 2559-3 c. 21,800 TN-2 18.8 9.5 3.4 4 4.3 15.4 Bone same individual 2559-1 This study: isotopes

Trou de Chaleux 2602 Late Glacial TC 20.1 4.7 3.2 39.1 14.1 Bone This study: isotopes

M. Germonpreet al. / Journal of Archaeological Science 36 (2009) 473490 477


6/18

results of herbivores from sites of a comparative age (20,00025,000 year old) were collated for assessment of the Pleniglacial

canid isotope results: Doln Vestonice, Milovice, and Bohuslavice(Ambrose, 1998), Laugerie-Haute Est (Drucker et al., 2003), andGoats Hole (Richards, 2000).

Samples were prepared and analyzed at the Research Laboratoryfor Archaeology and the History of Art (RLAHA), Oxford. Collagenwas extracted from each bone sample in the manner described byRichards and Hedges (1999). Samples were isotopically analyzed

using an automated Carlo Erba carbon and nitrogen elementalanalyzer coupled to a continuous flow isotope ratio-monitoring(PDZ Europa Geo 20/20) mass spectrometer. Each sample was run

in duplicate, with analytical errors of0.2& for both d13C and d15Nfor each analysis.

The genetic and isotopic analyses of the Goyet and the RussianPlain skulls are currently under investigation and the results ofthese analyses will be reported separately.

3. Results

3.1. Identification and grouping based on osteometry

The results of the PCA are given in Fig. 2and Table 3. The firsttwo principal components (PC1, PC2) explain 72.7% of the varianceamong all groups. The variables with the highest loadings on the

PC1are the two snoutwidth indices; the snout index has the lowestnegative loading. Variation along this axis ranges from a long,narrow snout at the left to a short, broad snout at the right. Mostfossil large canids have positive values while most of the recent

wolves show negative values on this axis. High values are seen inthe Eliseevich I dogs and two Chow chows from the recent archaicdog group. In the recent other dog groups, the lowest values are

seen in the Irish wolfhounds and Doberman Pinschers. From hereon they are grouped as recent other dogs with a slender snout. PC2is dominated by the tooth-row length index and the carnassiallength index. Variation along this axis corresponds to a relative

increase of the tooth-row and carnassial lengths. All fossil largecanids and prehistoric dogs have positive values on this compo-nent. The recent other dogs display both negative and positive

values depending on the breed. The Mastiffs, the Tibetan Mastiff,Rottweilers, and the Great Danes show generally low to very lowvalues on PC2. From here on they are grouped as recent other dogswith a short tooth-row. The range of the German Shepherds andMalinois overlaps largely with the one of the recent wolves. From

here on they are grouped as recent other dogs with wolf-like snout.The position of the Central Asian shepherd is outside the ranges ofall recent dogs, although quite close to the one of the recent archaic

dogs. As it appears isolated, it remains unclassified.The range of the prehistoric dogs falls completely outside the

ranges of the recent wolves and dogs. Mezin 5490, Mezhirich and

-4 -3 -2 -1 0 2 3 41

-3

-2

-1

0

1

2

3

Prin Comp 1

PrinComp2

Y+

XX Chow Chow

HuskyMalinoisGerman ShepherdDoberman PinscherIrish Wolfhound

Central Asian ShepherdGreat DaneMastiffRottweilerTibetan MastiffEliseevichi dogsEpipalaeolithic andMesolithic dogs

prehistoric canids

recent wolvesrecent archaic dogsrecent other dogs with wolf-like snoutrecent other dogs with short tooth rowrecent other dogs withslender snout

Y

Y

X

**

*

**

X

X

X

Y

Y

+

+

+

+

fossil large canidrecent wolves

recent young wolves recent zoo wolves

X

X

X

X

X

X

X

X

XX

X

X

X

X

Z

**

Y

X

Z

2

9

8

3

116

10

1

7

Fig. 2. Principal component analysis showing the first two principal components based on all indices (log 10-transformed). 1, Goyet; 2, Trou des Nutons; 3, Trou Balleux; 6, Mezin5469; 7, Mezin 5488; 8, Mezin 5490; 9, Mezhirich; 10, Avdeevo; 11, Yakutia.

Table 3

Principal component loadings and eigenvalues from the principal component

analysis of all indices (log10-transformed).

PC1 PC2

Eigenvalue 2.63 1.73

% Explained 43.85 28.81

% Cumulative 43.85 72.66

Eigenvectors

Carnassial length index 0.19 0.64Tooth-row length index 0.05 0.68

Snout index 0.44 0.28Greatest snout width index 0.59 0.01

Smallest snout width index 0.58 0.04

Braincase width index 0.31 0.22



7/18

Goyet fall in between the ranges of the prehistoric dogs, the recentarchaic dogs and the recent wolves. The other fossil canids aresituated inside the recent wolf range. The zoo wolves are at the

border of this range.The fossil large canid skulls from Trou de la Naulette and Grands

Malades have missing values and could not be included in the PCA.We used a DFA to assign the unclassified fossil large canids to

a classified group (recent archaic dogs, recent other dogs with shorttooth-row, recent other dogs with slender snout, recent other dogswith wolf-like snout, recent wolves, prehistoric dogs), on the basisof the log-transformed indices. Wilks lambda is very low implying

high case predictability (l 0.04, P< 0.0001). The DFA allowedclear separation of the canids: 92.9% of the cases are correctlyclassified. Table 4 gives the eigenvalues and eigenvectors of the first

two functions, which together account for 89.5% of the variation

among the groups. The canonical plot illustrates that the groupshave well-separated centroids (Fig. 3). Therefore the DFA is veryrobust, the variables selected are suitable and the groups can beconsidered as different. The Goyet, Mezin 5490 and Mezhirich

skulls have a probability of 0.99, 0.73 and 0.57, respectively, ofbelonging to the prehistoric dog group. The Avdeevo skull isattributed to the recent other dogs with wolf-like snout. All other

fossil large canids are predicted to be recent wolves. Interestingly,not one recent wolf is misclassified. The seven misclassified spec-imens are two huskies that are assigned to the prehistoric dogs, andfive German Shepherds that are attributed to the recent archaic

dogs (n 2), recent other dogs with slender snout (n 1) andrecent wolves (n 2). The unclassified Central Asian Shepherd isgrouped with the prehistoric dogs. The unclassified young wolvesare assigned to the recent wolves and the zoo wolves to the recent

other dogs with wolf-like snout.Based on the PCA and the DFA, the following skulls of the fossil

large canids are identified as prehistoric dogs: Goyet, Mezhirichand Mezin 5490, and are assigned to the prehistoric dog group. The

following canid skulls are interpreted as wolves: Trou des Nutons,Trou Balleux, Mezin 5469, Mezin 5488, and, as presumed, the skull

from Yakutia. They are further referred to as fossil wolves. Theattribution of the Avdeevo skull is unclear.

The results of the ANOVAs and on the KruskalWallis test on theindices are shown in Table 5a. All indices differ significantly amongthe large canid groups. As we are especially interested in the rela-

tionships between the prehistoric dogs, fossil wolves and recentwolves, ANOVAs were separately performed on these three groups,

Table 4

First canonical function and second canonical function observed from the

Discriminant Function Analysis based on all indices (log10-transformed).

Canon1 Canon2

Eigenvalue 4.12 1.88

% Explained 61.45 28.05

% Cumulative 61.45 89.50

Eigenvectors

Carnassial length index 35.04 17.49Tooth-row length index 33.64 15.98

Snout index 30.72 14.82Greatest snout width index 34.04 17.30

Smallest snout width index 3.23 17.83

Braincase width index 1.89 26.91

Doberman PinscherIrish WolfhoundGreat DaneMastiffRottweilerTibetan MastiffCentralAsian Shepherd

**

Y

X

Y+

X

X

fossil large canidsrecent wolvesrecent young wolvesrecent zoo wolvesEliseevichi dogs

Epipalaeolithic andMesolithic dogsChow ChowHuskyMalinoisGerman Shepherd

X

Y

Y

+++

+

++

Y

Y

*

*

*

*

*

X

X

X

X

XX

X

XX

X

XXX

X X XX

X

-39

-40

-41

-42

-43

-44

-45

-46

-38

2423 25 26 27 28 29 30 31 32 33 34 35

Canonical 1

Canon

ical2

group centroid.

.

..

.

.

.

Z

Z

1

98

102

3

7611

prehistoric canids

recent wolves

recent archaic dogs

recent other dogs with

wolf-like snout

recent other dogs with short

tooth row

recent other dogs with

slender snout

-36

-37

Fig. 3. Discriminant function analysis showing the first two discriminant functions performed on all groups using all indices (log 10-transformed); seeFig. 2for numbering of thespecimens.



8/18

followed by TukeyKramer HSD pairwise comparisons for thoseratios that are significantly different (Table 5b). The comparisonspoint out the following differences between the three groups. The

snout width indices differ significantly between the prehistoricdogs and those from both the fossil and recent wolves. Further-more, the carnassial length index and the tooth-row length indexdiffer significantly between the prehistoric dogs and the recent

wolves. Interestingly, the fossil wolves differ significantly from therecent wolves in one index: their greatest snout width index islarger than that of the recent wolves (Tables 5a, 5b). To account forthe potential influence of the paedomorphic morphology on the

braincase width index of the smaller Mesolithic and Epipalaeolithicdogs (cf.Wayne, 1986), the prehistoric dog group was split furtherinto two groups: one group containing the Palaeolithic dogs (all

having large skulls), and a second group with the smaller Epi-palaeolithic and Mesolithic dogs (Table 5c). The braincase widthindex of the Palaeolithic dogs differs significantly from both theones of the fossil and recent wolves. The Epipalaeolithic and

Mesolithic dogs differ for this index significantly from the threeother groups. The two fossil dog groups do not differ in the otherindices. However, the Palaeolithic dogs have snout width indicesthat are significantly larger than those of the recent and fossil

wolves. The carnassial length index differs significantly betweenthe Palaeolithic, Epipalaeolithic and Mesolithic dogs and the recentwolves (Table 5c).

The ANOVAs of the measurements among all canid groups aregiven inTable 6a. ANOVAs were also performed separately on thePalaeolithic dogs, the Epipalaeolithic and Mesolithic dogs, the fossilwolves and the recent wolves (Table 6b). Pairwise comparisons

show that the skull total length in the Palaeolithic dogs differssignificantly from those of the recent wolves and the Epi-palaeolithic and Mesolithic dogs. The viscerocranium length of the

Palaeolithic dogs differs significantly from that of the fossil andrecent wolves, and the Epipalaeolithic and Mesolithic dogs. Theircarnassial length differs from that of the Epipalaeolithic and

Mesolithic dogs but resembles that of the wolves. The fossil wolves

greatest snout width differs significantly from that of the recentwolves.

3.2. Ancient DNA analyses

All six large canid samples from Goyet cave and the singlesample from Trou des Nutons from which DNA was extractedyielded products of the expected size. Each sample resulted ina unique, reproducible sequence as described in Section2. Inter-

estingly, when compared to extant wolf and dog sequences avail-able from GenBank, all seven haplotypes found in the Pleistocenesamples were found to be unique and not described to date. Thisresult is remarkable when considering the large number of wolf

(w160) and particularly dog sequences (>1000 from almost allbreeds known today) available in Genbank. Since none of theancient sequences could be found to be identical to one of the

extensively genotyped dog breeds we assume them to representwolves rather than dogs. However, it is not possible to assignindividuals to either wolf or dog based only on mitochondrial DNAhaplotypes. This would only be possible using microsatellites,

which for technical reasons could not be applied here. Also forreasons of clarity the following phylogenetic comparisons onlyinclude wolf sequences since a phylogenetic tree including all thedog sequences would require several pages. In a neighbour-joining

tree using the software MEGA3.1 (Kumar et al., 2004) with Kimur-2-parameter corrected distances, five of the sequences, includingthe fossil wolf of Trou des Nutons, fall together with modern wolfsequences from Europe occurring from France across Italy and

Bulgaria to Greece (Fig. 4). However, it should be noted that thebootstrap support for this relationship is negligible as it is often thecase for intraspecific sequence comparisons. The same is true for

the basal position of the remaining two Pleistocene sequenceswithin Asian and European wolves. Since the interpretation of thisphylogenetic tree is delicate we also employed a haplotypenetwork displaying the genetic distance in terms of substitutional

steps separating the sequences of interest. The network shows thatall the Pleistocene sequences can be found in one half of thenetwork rather than being scattered across the complete network(Fig. 5).

3.3. Stable isotope analyses

The C/N ratios calculated forall of the samples in this studywerebetween 2.9 and 3.6, a range considered to be indicative of good

collagen preservation (Table 2)(Ambrose, 1990; DeNiro, 1985). Fullresults from the herbivore d13C and d15N values from each

Table 5b

Results of ANOVAs of the significant indices of selected groups

Index F P TukeyKramer

Greatest snout width 31.03


9/18

Late-glacial canid locality are given inTable 7. The isotope valuesrange from 22.8& to 17.8& in d13C and from 4.7& to 13.7& ind

15N (Fig. 6). Both the carbon and nitrogen isotope signatures of the

two fossil canid specimens from Trou des Nutons are almostidentical (Fig. 7). In contrast, approximately 2.5& separate the d13Cvalues of the two specimens from Trou du Frontal (Fig. 8). The fossil

canids from Goyet Cave have relatively similard13C values; howevertheir d15N signatures differ by up to 3.4& (Fig. 9). No obviouschronological pattern is present in the fossil canid isotope

signatures.In general the herbivores and canids have low d15N values which

are typical for fauna dating to the Late-glacial period ( Richards andHedges, 2003; Stevens and Hedges, 2004). It should be noted thatthe hares from Goyet Cave and the red deer from Trou de Chaleux

and Trou du Frontal have low d13C and high d15N values that aremore typical of Holocene specimens. These specimens were,however, found in contexts that are thought to be Late-glacial inage (based on radiocarbon dating). Direct dating of the specimens

would be required to investigate whether they are in fact Holoceneintrusions.

4. Discussion

Our results are consistent with the hypothesis that changes indog morphology compared to wolf morphology appeared ratherabruptly, were linked to the effects of domestication (see also theparagraphbelow on the genetic data) and that they became fixed inthe dog population. We have shown that the skull morphology of

a number of specimens is very similar and distinct from themorphology of wolves. These fossil skulls are interpreted as origi-nating from Palaeolithic dogs. In the PCA graph (Fig. 2) these dogsfall in between the ranges of the recent wolves, fossil wolves, recent

dogs and prehistoric dogs; in the DFA they are assigned to theprehistoric dogs (Figs. 2, 3). The oldest dog in our data set, with anAMS age of c. 31,700 BP, is from the Goyet cave (Belgium). The age ofthe skull implies that it could be derived from an Aurignacian

occupation. We therefore tentatively propose that the domestica-tion of the dog had already begun in the Aurignacian. Otherevidence for an early date for the domestication of the dog was

found in Chauvet cave, France. Here, occurring in the deepest partof the cave, a track of footprints from a large canid is associatedwith the one of a child (Garcia, 2005). Torch wipes made by this

child were dated at c. 26,000 BP. Based on the short length ofmedial fingers in the footprints the canid track was interpreted asbeing made by a large dog (Garcia, 2005). Genetic data also suggest

that the domestication of the wolf started before 15,000 BP (e.g.Lindblad-Toh et al., 2005; Vilaet al., 1999).

The Goyet skull is very similar to Eliseevich I dogs and to theEpigravettian Mezin 5490 and Mezhirich dog skulls, which areabout 18,000 years younger. The Mezin 5490 skull was classified by

Benecke (1987)as a captive wolf. A different approach involvingother reference groups, i.e. a more diverse set of measurements canpossibly explain the divergent interpretations. Pidoplichko (1998)interpreted the Mezin skull 5490 as from a dog.

Compared to wolves, ancient dogs exhibit a shorter and broadersnout (Lawrence, 1967; Olsen, 1985; Sablin and Khlopachev, 2002).All Palaeolithic dogs in our study conform to this pattern. Theirviscerocranium length is significantly shorterand their snout width

indices are significantly larger than those of the fossil and recent

wolves (Figs. 11 and 12, Table 5c). Furthermore, the mean totallength of their skull is smaller than that in recent and fossil wolves,although not significantly compared to the latter group. In addition,

the Palaeolithic dogs have a braincase width index that is largerthan that of fossil and recent wolves (Figs. 12, 13,Table 5c).Morey(1992)also found that postglacial dogs have proportionally widerpalates and braincases than the wild Canis. According toClutton-

Brock (1997)evolutionary reduction in the size of the teeth tookplace at a slower rate than the shortening of the snout. In thePalaeolithic dogs, absolute and relative carnassial length reductioncan indeed not be observed (Tables 5c and 6b).

The characteristics of the snout of the Palaeolithic dogs implypowerful jaws with well-developed carnassials. Their skull size

Table 6a

Results of ANOVAs of the raw measurements.

Measurement F P Palaeolithic

dogs

n5 (a1)

Epipalaeolithic

and Mesolithic

dogs n 3 (a2)

Fossil

wolves

n5 (b)

Recent

wolves

n46 (c)

Recent

archaic

dogs

n18 (d)

Recent

other

dogs with

wolf-like

snout

n17 (e)

Recent other dogs

with short

tooth-row

n11 (f)

Recent other

dogs with

slender

snout

n 6 (g)

TL (1) 40.22


10/18

suggests that these first dogs were large animals, as large as recentlarge dog breeds (Table 6a). They may have been used for helpingwith the tracking, hunting or transport of large, ice-age game,possibly mammoths on the Russian Plain (Sablin, 2002; Germonpre

et al., in press) or used as pack animals (cf. Balikci, 1970; Morey,1986). In the PCA and DFA plots, the Palaeolithic dogs are situatednear theCentral Asian Shepherd dog, which wasassigned in theDFAto the prehistoric dog group. This suggests that their skull shape

resembles that of the latter breed which was originally used asa flock guardian and as a protector against predators such as bears,striped hyenas and wolves (Labunsky, 1994). The latter property in

a dog could also have been useful for Palaeolithic people.As demonstrated above, the Palaeolithic dogs in our data set are

very uniform in their skull shape. Even the Goyet dog, with an ageof c. 31,700 BP, is not intermediate in form between the fossil

wolves and the prehistoric dogs, but conforms to the configurationof the other Palaeolithic dogs, which are approximately 18,000years younger. The abrupt appearanceof a dog, much older than the

Eliseevich I dogs, the oldest recognized dogs so far, suggest that thedomestication process must have been quite rapid (cf. Crockford,2000a). Once the Palaeolithic dogs were established, their skullmorphology seems to have remained stable. Other early dogs (e.g.

Jomon dogs) equally display a remarkable similarity in skull shapethat persevered for thousands of years (Crockford, 200 0b; Shige-hara and Hongo, 2000).

The skulls from Trou Baileux, Trou des Nutons, Mezin 5469,Mezin 5488 and Yakutia are considered to be from fossil wolves.They differ in their greatest snout width and also in the greatestsnout width index from the recent wolves (Tables 5b, 6b). A

similar trend was discovered in North American fossil wolves.Eastern Beringian wolves tend to have short broad palates, prob-ably adapted for producing relatively large bite forces suggesting

the killing of bulky prey, such as bison and horse (Leonard et al.,2007).

European haplotype

Asian haplotype

Eurasian haplotype

Belgian fossil haplotype

z coyote

G-3G-6

TN-1

G-4G-5

G-2G-1

z coyote

Fig. 4. Neighbour-joining tree of ancient Belgian large canids and recent wolves based

on 57 bp of mitochondrial control region. Ancient Belgian haplotypes are shown as G-

1 to G-6 (Goyet) and TN-1 (Trou des Nutons), respectively. Due to the short

sequence length, bootstrap values are low for all nodes. AMS age of G-5, c. 13,700 BP;G-6, c. 24,800 BP; TN-1, c. 21,800 BP (Table 2).

z coyote

TN-1

G-5

G-4

G-3

G-1

G-2

G-6

European haplotype

Asian haplotype

Eurasian haplotype

Belgian fossil haplotype

z coyote

Fig. 5. Median-joining haplotype network depicting the phylogenetic relationships

between recent European and Asian wolves as well as ancient Belgian large canids. The

size of the circles indicates the number of individuals carrying a particular haplotype.

SeeFig. 4for abbreviations.

Table 6b

Results of ANOVAs of the raw measurements of selected groups

Measurement F P TukeyKramer

TL (1) 41.70


11/18

Table 7

Results of herbivore d13C and d15N with %C, %N and C/N ratios from Goyet Cave, Trou de Chaleux and Trou du Frontal.

Sample Species Site code %C %N d13C d15N C:N AMS code AMS date

Goyet Cave Horizon 1

A/GOY/B/16 Capra ibex DP2778/54 38.2 13.5 19.8 2.9 3.3

A/GOY/B/18 Cervus elaphus DP2779/3 39.4 14.8 21.7 5.4 3.1

A/GOY/B/19 C. elaphus DP2781/2 32.8 11.6 19.3 5.5 3.3

A/GOY/B/1 Equussp. DP2832/2 40.2 14.7 20.4 7.3 3.2A/GOY/B/2 Equussp. DP2832/3 41.4 15.1 20.2 4.1 3.2

A/GOY/B/3 Equussp. DP2832/6 34.3 12.6 20.5 7.0 3.2

A/GOY/B/4 Equussp. DP2832/1 36.0 13.2 20.8 5.3 3.2

A/GOY/B/5 Equussp. DP2813/8 35.0 12.4 21.1 5.8 3.3

A/GOY/B/6 Equussp. DP2813/1 27.1 10.6 20.9 1.7 3.0

A/GOY/B/7 Equussp. DP2813/4 40.1 14.4 20.2 4.5 3.3

A/GOY/B/8 Equussp. DP2813/3 40.7 14.8 20.7 5.4 3.2

A/GOY/B/9 Equussp. DP2813/6 Failed Failed Failed Failed Failed

A/GOY/B/10 Equussp. DP2813/11 26.2 9.6 21.3 5.3 3.2

A/GOY/B/11 Ovibos moschatus DP2783/1 34.5 12.9 19.3 3.9 3.1

A/GOY/B/14 Rangifer tarandus DP2778/41 40.4 14.7 18.0 6.2 3.2A/GOY/B/15 R. tarandus DP2778/49 30.3 11.5 18.6 0.4 2.9

A/GOY/B/17 R. tarandus DP2778/52 40.2 14.5 19.0 3.9 3.2

Goyet Cave Horizon 2

A/GOY/B/27 Bison priscus DP2769/1 34.5 12.7 20.3 3.6 3.2

A/GOY/B/28 B. priscus DP2824/4 Failed Failed Failed Failed 2.8

A/GOY/B/29 B. priscus DP 2769/6 41.1 15.0 20.1 7.1 3.2

A/GOY/B/31 Equussp. DP2809 25.1 9.7 20.8 3.0 3.0

A/GOY/B/32 Equussp. DP2809 31.5 11.6 21.0 6.0 3.2

A/GOY/B/33 Equussp. DP2809 39.8 14.3

20.8 6.4 3.2A/GOY/B/34 Lepussp. NO CODE 42.8 15.4 21.6 3.9 3.2

A/GOY/B/41 Lepussp. NO CODE 43.6 15.4 23.5 4.7 3.3A/GOY/B/30 O. moschatus DP2770/3 34.0 12.5 19.6 4.0 3.2

A/GOY/B/35 R. tarandus DP2768 41.3 15.0 19.3 2.8 3.2

A/GOY/B/36 R. tarandus DP2768 42.0 15.2 19.4 2.6 3.2

Trou de Chaleux

A/CX/B/1 B. priscus DP2591/17 27.5 11.0 19.7 1.7 2.9

A/CX/B/2 B. priscus DP2591/14 Failed Failed Failed Failed FailedA/CX/B/3 C. elaphus DP2591 32.5 11.0 23.5 5.6 3.5

A/CX/B/4 C. elaphus DP2591 Failed Failed Failed Failed Failed

A/CX/B/13 C. elaphus DP2596 30.1 10.5 22.6 3.3 3.3A/CX/B/27 Equussp. DP2297 29.0 11.1 20.3 0.4 3.1

A/CX/B/28 Equussp. DP2298 19.7 6.3 21.0 2.4 3.6

A/CX/B/29 Equussp. DP2298 38.0 13.5 20.9 1.4 3.3

A/CX/B/30 Equussp. DP2297 42.4 15.2 20.7 2.2 3.3

A/CX/B/31 Equussp. DP2298 36.6 12.6 20.8 2.3 3.4

A/CX/B/32 Equussp. DP2298 38.5 13.7 20.9 3.3 3.3A/CX/B/33 Equussp. DP2298 39.3 14.1 20.6 1.9 3.3

A/CX/B/34 Equussp. DP2298 42.5 15.3 20.8 1.9 3.2

A/CX/B/35 Equussp. DP2298 40.4 14.4 21.0 1.6 3.3A/CX/B/36 Equussp. DP2298 42.6 15.4 21.1 2.1 3.2

A/CX/B/37 Equussp. DP2298 50.2 17.8 20.9 1.9 3.3

A/CX/B/38 Equussp. DP2298 Failed Failed Failed Failed Failed

A/CX/B/39 Equussp. DP2298 Failed Failed Failed Failed Failed

A/CX/B/40 Equussp. DP2298 36.5 13.1 20.9 1.3 3.2

A/CX/B/41 Equussp. DP2298 Failed Failed Failed Failed 2.5

A/CX/B/42 Equussp. DP2298 21.1 7.7 21.0 0.6 3.2

A/CX/B/43 Equussp. DP2298 19.1 6.8 21.1 2.3 3.3

A/CX/B/45 Equussp. DP2342 41.9 15.1 20.7 1.3 3.2 OxA-3633 12,880 100

A/CX/B/46 Equussp. DP2342 36.9 13.2 21.1 2.1 3.3 OxA-3632 12,790 100

A/CX/B/14 Lepussp. DP2623 39.7 14.2 20.7 1.6 3.3

A/CX/B/15 Lepussp. DP2623 43.6 15.9 19.8 6.6 3.2A/CX/B/16 Lepussp. DP2623 42.3 15.0 21.1 0.9 3.3

A/CX/B/17 Lepussp. DP2623 42.0 15.0 21.0 1.0 3.3

A/CX/B/5 O. moschatus DP2595/15 Failed Failed Failed Failed FailedA/CX/B/6 O. moschatus DP2595/23 25.0 9.0 19.7 4.2 3.2

A/CX/B/7 O. moschatus DP2595/19 26.5 9.7 19.9 2.9 3.2


A/CX/B/9 O. moschatus DP2595/17 Failed Failed Failed Failed Failed

A/CX/B/10 O. moschatus DP2595/16 Failed Failed Failed Failed 3.7


A/CX/B/44 O. moschatus DP2594/33 37.8 13.6 19.3 4.3 3.2 OxA-4192 12,860 140

A/CX/B/12 R. tarandus DP2599 43.4 15.2 19.1 2.6 3.3

Trou du Frontal

A/FRT/B/1 Bos primigenius DP2456/1 40.5 14.8 20.1 3.7 3.2

A/FRT/B/12 C. elaphus DP2448 42.0 15.3 22.1 3.4 3.2

A/FRT/B/2 Equussp. DP2449 42.0 14.9 20.8 1.7 3.3

A/FRT/B/3 Equussp. DP2449 44.7 16.1 21.0 1.7 3.2

A/FRT/B/9 Equussp. DP2450 43.4 15.6 20.3 2.1 3.2 OxA-4197 12,800 130

A/FRT/B/7 Lepussp. DP2564 42.5 14.7 21.9 0.3 3.4

A/FRT/B/8 Lepussp. DP2564 44.4 16.0 19.6 1.0 3.2


12/18

The Gravettian Avdeevo specimen is, like the two zoo wolves in

the DFA, assigned to the group of the recent other dogs with wolf-like snout. It is possible that it also was kept in captivity.

The Trou de la Naulette and the Grandes Malades skulls havemissing values and cannot be identified.

The seven mtDNA sequences of the Belgian fossil large canidswere compared to the mtDNA haplotypes of recent dog breeds andrecent wolves. One of the Belgian specimens, the Trou des Nutons

skull, was identified on an osteometric basis to be a fossil wolf. Theancient DNA sequences are unfortunately too short to allow

rigorous phylogenetic analyses. Nevertheless, a few conclusions canbe drawn.

First, all Pleistocene samples yielded unique sequences. Thisindicates that the Belgian large canids carried a substantial amountof genetic diversity. Since dogs were domesticated from graywolves, ultimately the first dogs would have carried a wolf-like

genetic sequence, and hence will be not identifiable genetically asthe first dogs. Only after isolated breeding, it is possible that certaingenotypes in the Palaeolithic dogs drifted to high frequencies and

-22 -21 -20 -19 -18

0

2

4

6

8

10.

13C

15N

-23

X

.

+Y

Goyet Cave CanidsTrou du Frontal Canids

X

. + YBos/Bison

Capra ibexCervus elaphusEquus sp.

Lepus sp.Ovibos moschatusRangifer tarandusTrou de Chaleux Canid

Fig. 6. Stable carbon and nitrogen isotope signatures of fossil large canids and asso-

ciated herbivores from the Late-Glacial Belgian sites of Trou de Chaleux, Trou du

Frontal and Goyet cave.

-22 -20-21 -19 -18

0

2

4

6

8

10

X

X

X

XX

13C

15N

-23

Y

Canis lupus

LepusEquus sp.

Mammuthus primigenius

Rangifer tarandus

12

Y

Y+

+

Diet of TN-1

Diet of TN-2

TN-2

TN-1

Fig. 7. Carbon and nitrogen isotope signatures of fossil wolves from Trou des Nutons

and contemporary herbivores from Doln Vestonice, Milovice, and Bohuslavice

(Ambrose, 1998), Laugerie-Haute Est (Drucker et al., 2003), and Goats Hole (Richards,

2000). Rectangular boxes represent possible isotopic values of average prey consumed

by the wolves using the range of isotopic fractionations observed by Bocherens and

Drucker (2003). TN-1, wolf with AMS age of c. 21,800 BP; TN-2, both specimens arefrom the same wolf skeleton (Table 2).

X

Y

+

-23 -22 -21 -20 -19 -18

10

9

8

7

6

5

4

3

2

1

0

G-7

G-8

G-2G-5

Diet of G-5 Diet of G-2Diet of G-8

Diet of G-7

Fossil large canid

Equus sp.

Cervus elaphus

Rangifer tarandus

Bison priscus

Ovibus moschatus

Capra ibex

Lepus sp.

X

+

Y

13C

15N

G-1

Diet of G-1

Fig. 9. Carbon and nitrogen isotope signatures of fossil wolves and herbivores from

Goyet Cave. Error bars, standard deviation of mean. Rectangular boxes represent

possible isotopic values of average prey consumed by each wolf using the range of

isotopic fractionations observed by Bocherens and Drucker (2003). See Table 2 forabbreviations, G-5 with an AMS age of c. 13,700 BP.

-18-19-20-21-22-23

10

8

6

4

2

0

Fossil large canid

Equus sp.

Cervus elaphus

Bos primigenius

Lepus sp.

TF-1TF-2

Diet of 1 Diet of 2

Y Y

1

3

5

7

9

13C

X

X

15N

Fig. 8. Carbon and nitrogen isotope signatures of fossil canids and herbivores from

Trou de Frontal. Error bars, standard deviation of mean. Rectangular boxes representpossible isotopic values of average prey consumed by each canid using the range of

isotopic fractionations observed by Bocherens and Drucker (2003). See Table 2 for

abbreviations.



13/18

might therefore be distinguishable from those of the source wolfpopulation. Thus one would only expect to see a differentiation ofdogs and wolves after several thousands of years due to thebottleneck caused by selective breeding during early domestica-

tion. At the same time certain (e.g. morphological) traits wereprobably expressed and selected for, or just arose by drift. After thisinitial phase of domestication a relaxation of constraints occurred.Wolves after the domestication (i.e. dogs) were insulated from the

full force of negative selection because humans cared for them, e.g.providing food and physical protection therefore (slightly)disadvantageous traits could still survive and produce offspring.

These individuals in their natural environment would probably nothave contributed to the next generations gene pool. Later, expo-nential population growth increased genetic diversity to the highlevels that are observed in dogs today. Since the domestication ofdogs is, in evolutionary timescales, a rather recent event, the line-

ages of wolves and dogs have not separated yet, and therefore donot allow rigorous identification of the analyzed specimens. Giventhe proposed timescale for the dog domestication of only a few tenthousands of years and the mtDNA mutation rate, this is not

unexpected. However, due to extensive breeding in the last coupleof hundred years, the soaring population size of dogs has provideda sufficiently large genetic background to accumulate relativelyhigh levels of mtDNA diversity. Usually this genetic diversity is

centred around the stock haplotype, which was initially selected for

or drifted to higher frequencies, and results in a star-like patternwhen depicted in a phylogenetic haplotype network. Given the

extensively typed mtDNA gene pool of dogs, it is highly unlikelythat the yet unknown mtDNA haplotypes in every single Belgianspecimen represent Palaeolithic dogs. We interpret the fact thatnone of the studied ancient canid sequences can be found in the

extensively typed gene pool of modern dog breeds as evidence thatthe investigated Belgian specimens do not represent Palaeolithicdogs. The seven unique haplotypes thus suggests that the sampled

fossils represent ancient wolf lineages that either became lost untilnow, or have not been observed so far in extant wolves. This is alsoreinforced by the morphometric results, e.g. of the Trou des Nutonsspecimen which clearly has a wolf-like morphology. It should be

noted that the microsatellite analyses necessary for identification

cannot be performed on the investigated specimens due to theirpoor DNA preservation.

Second, the mtDNA sequences found have not been describedfrom modern wolves, suggesting that the genetic diversity ofmodern wolves is reduced compared to their Pleistocene ancestors,either due to climatic changes (Shapiro et al., 2004) or due to

human impact on modern wolf populations (Vilaet al., 1999).Third, as in modern wolves, there is little evidence for phylo-

geographic structure in the Pleistocene wolves since they do notform a homogenous genetic group. However, there is some

evidence of geographic cohesion, at least with respect to theBelgian wolves, as their sequences are not scattered throughoutthe full network, but concentrated in one half of it ( Fig. 5). In theneighbour-joining tree, the fossil wolf of Trou des Nutons, together

with five other sequences, occurs together with modern wolfsequences from Europe (Fig. 4).

We conclude that although it is possible that a number ofgenotypes of the Palaeolithic dogs were lost during the last 30,000

years, it is more likely that the Belgian haplotypes represent fossilwolves. The dog population has grown exponentially which meansthat almost every haplotype that arose from a new mutation, wascarried to the next generation. In wolves, however, the population

size declined due to human impact and/or climatic changes, whichcaused a substantial loss of genetic diversity. The far most feasible

explanation for the observed data is that the unique Belgianhaplotypes represent ancient diversity in wolves, which now nolonger exists, rather than newly arisen dog haplotypes that wentlost during the massive population expansion of the dog.

The mean d13C and d15N values of the Late-glacial wolves fromBelgium (as determined by morphometric and DNA analyses) are0.8& and 4.1& higher, respectively, than those of associated Late-

glacial Belgian herbivores (Fig. 6). These differences correspondwell with the observed trophic enrichment of 0.81.3& for carbonand 35& for nitrogen observed between herbivores and modernwolves (Bocherens and Drucker, 2003).

Although it is tempting to apply a mixing model such as thatproposed byPhillips and Koch (2002)to determine the proportionof different prey consumed at each site, the limited amount of

isotope data available from potential prey makes the application ofsuch models to this data set unwise. Initial comparisons betweenthe Late-glacial fossil wolf and herbivore isotope signatures arebased on site-related average values as differences in herbivored

15N values are visible between sites despite their close geographic

proximity. This is most noticeable for horse at Trou de Chaleux andGoyet Cave and is likely to relate to differences in the averageisotope composition of their diets due to a climatic gradient withinthe region or due to chronological difference between the sites.

These differences in herbivored15N values are currently the focus offurther investigation through isotopic analysis and radiocarbondating (Stevens et al., in press). Wolves often occupy large homeranges and move around the landscape. Thus, the Late-glacial

wolves and herbivores arealso compared to average species isotope

signatures regardless of site (Fig. 6). This latter approach alsofacilitates comparison with the Pleniglacial wolf isotope signaturesfrom Trou de Nutons (Fig. 7).

At Goyet, the d15N of the five wolves are variable, ranging from5.5& to 8.9&whereas thed13C vary by less than 1& (Fig. 9). Usingthe pattern of trophic level enrichment observed in modern

ecosystem (Bocherens and Drucker, 2003), we can postulate thatspecimen G-7 and G-5 primarily consumed horses. For specimenG-8 and G-2 the calculated isotope values of the average preyconsumed overlaps with more than one species. Both horse and

bison appear to have been important prey for these two specimens.For specimen G-1 the calculated isotope values of the average preyconsumed overlaps very slightly with that of the horse, however,this species does not seem to be the main prey consumed. The

calculated isotope values of the average prey consumed is notconsistent with any of the herbivore analyzed from Goyet cave, but

Fossil large canid

Equus sp.

Cervus elaphus

Rangifer tarandus

-24 -23 -22 -21 -20 -19 -18

Diet of large canid

Bison priscus

Ovibus moschatus

Lepus sp.

5

4

3

2

1

0

-1

13C

Y

Y

6

X

X

15N

Fig. 10. Carbon and nitrogen isotope signatures of fossil canid and herbivores from

Trou de Chaleux. Error bars, standard deviation of mean. Rectangular box represents

possible isotopic values of average prey consumed by the canid using the range of

isotopic fractionations observed byBocherens and Drucker (2003).



14/18

is consistent with those of horse and hare at Trou de Chaleux andTrou du Frontal (Fig. 6). The calculated d13C of the average preyconsumed for all canids from Goyet cave are 12& lower than

average reindeer d13C, suggesting reindeer was not a major input in

their diet. The variation in wolf and herbivore d15

N at Goyet cavemay be chronologically related as faunal d15N has been shown torapidly fluctuate during the late Pleistoceneearly Holocene tran-

sition (Drucker et al., 2003; Richards and Hedges, 2003; Stevensand Hedges, 2004; Stevens et al., 2008).

At Trou de Chaleux the d15N of the large canid is verylowand the

associated herbivore isotope signatures are also very low (Figs. 6,10). The calculated range of isotope values of the average preyconsumed by the canid partially overlaps with the isotope values oftwo species, horse and hare, suggesting these may be major inputs

in the canids diet. With significantly higher isotope signatures thanthe calculated average prey, musk ox is unlikely to have beenconsumed by the Chaleux canid.

At Trou du Frontal the two canids have similar d15N values but

their d

13

C values differ by 2.4& (Figs. 6, 8). The majority of theHolocene faunal d13C values are around 2& lower than Late

Pleistocened13C values due to a changein plant d13C values over time(Hedges et al.,2005; Richardsand Hedges, 2003;Stevens and Hedges,2004). Wolfd13C values in particular have been shown to track plant

and atmospheric d13C(Bump et al., 2007). Neither of the two canids

from Trou du Frontal have been radiocarbon dated, and with suchdivergent carbon isotope signature one could speculate that spec-imen TF-2 is Late Glacial in age, and specimenTF-1is olderor younger

in age. The calculated range of isotope values of the average preyconsumed by the wolves is not consistent with any of the herbivorespecies analyzed. This is due to the limited herbivore isotope data

available from this site. However, the calculated range of isotopevalues of the average prey consumed by specimen TF-1 is closest tothe isotopic signature of the red deer from Trou du Frontal, andoverlaps with the red deer signatures at Goyet Cave and Trou de

Chaleux. As previously discussed these red deer specimens haveunusually isotopic signatures for a Late-glacial context and are moretypical of specimen that date to theHolocene. The calculated range ofisotope values of the average prey consumed by specimen TF-2 is

closest to that of theBos primigeniusfrom Trou du Frontal and over-laps with the isotopic signatures ofBison priscusat Goyet Cave.

Fig. 11. Dorsal view of the skulls from (a) Goyet (dog); (b) Trou Balleux (wolf); (c) Trou des Nutons (wolf), showing the relative wide braincase of the Goyet dog.



15/18


16/18

appear to be consuming horse and/or large bovids. Although there

is substantial genetic diversity in the Belgian wolves, they appear tobe consuming relatively similar diets. It is interesting to note thatthe fossil eastern Beringian wolves preyed mainly on horse and

bison (Leonard et al., 2007).The diet of Palaeolithic humans is thought to have broadened

during the mid/late Upper Palaeolithic to include freshwater

resources (Richards et al., 2001). During the earliest stages ofdomestication we might expect the diet of wolves to mimic thoseofhumans either due to feeding or scavenging and that these changesmight be detectable prior to genetic and morphological changes.

Some evidence to support this expectation comes from theMagdalenian site of Saint-Germain-la-Riviere (France). Isotopicanalysis of two large canids (identified as wolves by the authors)has shown that they, like the human from this site, consumed both

terrestrial herbivores and marine fish (anadromous salmon)(Drucker and Henry-Gambier, 2005). It is unlikely, however, that

these large canid skeletal remains (an ulna with a radiocarbon age

of 14,22016514

C BP, and a mandible from a layer dating between15,300 410 and 16,89013014C BP) could be conclusively iden-tified as wolves, or primitive dogs based on the morphology of the

elements available. A taphonomic study of the fish remains fromTrou de Chaleuxand Trou de Frontal indicates that the Magdalenianpeople at these sites exploited medium- and large-sized salmonids,

burbots and cyprinids, probably during the spawning season (VanNeer et al., 2007). The lack of evidence for either freshwater oranadromous fish consumption by the Belgian fossil wolves indi-cates their diets were not tracking those of the Upper Paleolithic

humans, possibly inferring they were not closely interacting.

5. Conclusion

On the basis of a PCA and a DFA, the fossil large canids fromGoyet, Mezhirich and Mezin (5490) are, together with those of

Fig.13. Ventral view of the skulls from (a) Goyet (dog); (b) Trou Balleux (wolf); (c) Trou des Nutons (wolf), showing the relative wide snout and large carnassial of the Goyet dog.



17/18

Eliseevich I, identified as Palaeolithic dogs. The antiquity of theGoyet skull suggests that dog domestication had possibly already

begun during the Aurignacian. The mtDNA of the Belgian fossilwolves shows a large amount of genetic diversity that has not beendescribed for modern wolves. Therefore, it is possible that modern

genetic diversity is less among recent wolves relative to their fossilconspecifics. This conclusion is supported by genetic data fromextant wolves which also indicate a reduction of wolf populationsat some point in the past (Vilaet al., 1999). The trophic interpre-

tation of the isotope data available from the Belgian fossil wolvessuggests that they relied mainly on horse and possibly large bovids;musk ox, reindeer, freshwater and anadromous fish were appar-

ently not very important prey animals. Thus the diets of the Belgianfossil wolves do not appear to be mimicking the increased dietbreath (and particularly the consumption of freshwater/anadro-mous fish) observed for mid/late Upper Palaeolithic humans.

Furtheranalyses of osteometric results, stable isotopes and longerancient DNA sequences, including additional loci, from other fossillarge canids would be necessary to shed more light on the earlydomesticationof thedog, and therelationship of fossilwolves both to

their modern counterparts and to the prehistoric and recent dog.

Acknowledgements

This work has been supported by a Belgian Federal Science

Policy Office fellowship (to M. Sablin), and the European-fundedgrants Access to Belgium Collections (ABC) and SYNTHESYS anda NERC Studentship (NER/S/A/2000/03522) (to R. Stevens). Wethank D. Ivanoff (PM NASU), M.V. Sotnikova (GIN RAS), G. Lenglet

(RBINS), E. Poty (ULg), V. Moiseyev (MAE RAS), E. Gillissen and P.Mergen (Royal Museum for Central Africa, Brussels). Anne Wautersis thanked for her help with the figures. We are grateful to threeanonymous reviewers for their helpful comments.

References

Altuna, J., Baldeon, A., Mariezkurrena, K., 1985. Cazadores magdalenienses en laCueve de Erralla (Cestona, Pais Vasco). Munibe 37, 1206.Ambrose, S.H., 1990. Preparation and characterization of bone and tooth collagen

for isotopic analysis. Journal of Archaeological Science 17, 431451.Ambrose, S.H., 1998. Prospects for stable isotopic analysis of Later Pleistocene

Hominid diets in West Asia and Europe. In: Akazawa, T., Aoki, K., Bar-Yosef, O.(Eds.), Neandertals and Modern Humans in Western Asia. Plenum Press, NewYork, pp. 277294.

Ambrose, S.H., Norr, L., 1993. Experimental evidence for the relationship of thecarbon isotope ratios of whole diet and dietary protein to those of bone col-lagene and carbonate. In: Lambert, J.B., Grupe, G. (Eds.), Prehistoric HumanBone Archaeology at the Molecular Level. Springer Verlag, Berlin, pp. 137.

Atchley, W.R., Gaskin, C.T., Anderson, D., 1976. Statistical properties of ratios I.Systematic Zoology 25, 137148.

Balikci, A., 1970. The Netsilik Eskimo. Waveland Press, Prospect Heights, Illinois.Bandelt, H.J., Forster, P., Rohl, A., 1999. Median-joining networks for inferring

intraspecific phylogenies. Molecular Biology and Evolution 16, 3748.Benecke, N., 1987. Studies on early dog remains from Northern Europe. Journal of

Archaeological Science 14, 3149.

Bocherens, H., Pacaud, G., Lazarev, P.A., Mariotti, A., 1996. Stable isotope abundances(C-13, N-15) in collagen and soft tissues from Pleistocene mammals fromYakutia: implications for the palaeobiology of the Mammoth Steppe. Palae-ogeography, Palaeoclimatology Palaeoecology 126, 3144.

Bocherens, H., Drucker, D., 2003. Trophic level isotopic enrichment of carbon andnitrogen in bone collagen: case studies from recent and ancient terrestrialecosystems. International Journal of Osteoarchaeology 13, 4653.

Bump, J.K., Fox-Dobbs, K., Bada, J.L., Koch, P.L., Peterson, R.O., Vucetich, J.A., 2007.Stable isotopes, ecological integration and environmental change: wolvesrecord atmospheric carbon isotope trend better than tree rings. Proceedings ofthe Royal Society B Biological Sciences 274, 24712480.

Chaix, L., 2000. A Preboreal dog from the Northern Alps (Savoie, France). In:Crockford, S.J. (Ed.), Dogs Through Time: An Archaeological Perspective,Proceedings of the 1st ICAZ Symposium on the History of the Domestic Dog.BAR International Series 889, pp. 4959.

Charles, R., 1998. Late Magdalenian Chronology and faunal exploitation in theNorth-Western Ardennes. In: BAR International Series 737.

Clutton-Brock, J., 1997. Origins of the dog: domestication and early history. In:Serpell, J. (Ed.), Domestic Dog: Its Evolution, Behavior and Interactions withPeople. Cambridge University Press, Cambridge, pp. 219.

Corrucini, R.S., 1987. Shape in morphometrics: comparative analyses. AmericanJournal of Physical Anthropology 73, 289303.

Crockford, S.J., 2000a. Dog evolution: a role for thyroid hormone physiology indomestication changes. In: Crockford, S.J. (Ed.), Dogs Through Time: AnArchaeological Perspective, Proceedings of the 1st ICAZ Symposium on theHistory of the Domestic Dog. BAR International Series 889, pp. 295312.

Crockford, S.J., 2000b. A commentary on dog evolution: regional variation, breeddevelopment and hybridization with wolves. In: Crockford, S.J. (Ed.), DogsThrough Time: an Archaeological Perspective, Proceedings of the 1st ICAZSymposium on the History of the Domestic Dog. BAR International Series 889,

pp. 295312.Dalen, L., Nystrom, V., Valdiosera, C., Germonpre, M., Sablin, M., Turner, E.,

Angerbjorn, A., Arsuaga, J.L., Gotherstrom, A., 2007. Ancient DNA reveals lack ofpostglacial habitat tracking in the arctic fox. Proceedings of the NationalAcademy of Sciences of the United States of America 104, 67266729.

Davis, S.J.M., Valla, F.R., 1978. Evidence for domestication of the dog 12,000 yearsago in the Natufian of Israel. Nature 276, 608610.

Degerbl, M., 1961. On a find of a Preboreal domestic dog ( Canis familiarisL.) fromStar Carr, Yorkshire, with remarks on other Mesolithic dogs. Proceedings of thePrehistoric Society 27, 3555.

DeNiro, M.J., 1985. Postmortem preservation and alteration of in-vivo bonecollagen isotope ratios in relation to palaeodietary reconstruction. Nature317, 806809.

Depaepe, P., 1988. Hulsonniaux (Houyet, Nr): Trou Balleux. Archeologie. Chroniquesemestrielle pour larcheologie en Belgique 2, 150.

Dewez, M., 1987. Le Paleolithique superieur recent dans les grottes de Belgique.Publications dHistoire de lArt et dArcheologie de lUniversite Catholique deLouvain 57, 1466.

Dikov, N.N., 1996. The Ushki sites, Kamchatka Peninsula. In: West, F.H. (Ed.),American Beginnings. The Prehistory and Paleoecology of Beringia. Universityof Chicago Press, Chicago, pp. 244250.

Drucker, D.G., Bocherens, H., Billiou, D., 2003. Evidence for shifting environmentalconditions in Southwestern France from 33,000 to 15,000 years ago derivedfrom carbon-13 and nitrogen-15 natural abundances in collagen of largeherbivores. Earth and Planetary Science Letters 216, 163173.

Drucker, D.G., Henry-Gambier, D., 2005. Determination of the dietary habits ofa Magdalenian women from St Germain-la-Riviere in southwestern Franceusing stable isotopes. Journal of Human Evolution 49, 1935.

Dupont, E., 1873. Lhomme pendant les ages de la pierre dans les environs deDinant-sur-Meuse. C. Muquardt, Bruxelles.

Fiedel, S.J., 2005. Mans best friend mammoths worst enemy? A speculative essayon the role of dogs in Paleoindian colonization and megafaunal extinction.World Archaeology 37, 1125.

Fry, B., Sherr, E.M., 1984.d13C measurements as indicators of carbon flow in marineand freshwater ecosystems. Contributions in Marine Science 27, 1347.

Garcia, M.A., 2005. Ichnologie generale de la grotte Chauvet. Bulletin de la Socie teprehistorique franaise 102, 103108.

Germonpre, M., 1996. Preliminary results on the mammals of the Magdalenianupper horizon of Goyet (Belgium). Notae Praehistoricae 16, 7585.Germonpre, M., 1997. AMS dating of the Upper Magdalenian horizon 1 at Goyet

Cave. Bulletin de L lnstitut royal des Sciences Naturelles de Belgique (Sciencesde la Terre) 67, 167182.

Germonpre, M., Hamalainen, R., 2007. Fossil bear bones in the Belgian UpperPalaeolithic: the possibility of a proto-bear ceremonialism. Arctic Anthropology44, 130.

Germonpre, M., Sablin, M., Khlopachev, G.A., Grigoreiva, G.V., Possible evidence ofmammoth hunting during the Epigravettian at Yudinovo, Russian Plain. Journalof Anthropological Archaeology, in press (doi:10.1016/j.jaa.2008.07.003).

Gvozdover, M., 1995. Art of the mammoth hunters. The finds from Avdeevo, OxbowMonograph 49.

Harcourt, R.A., 1974. The dog in prehistoric and early historic Britain. Journal ofArchaeological Science 1, 151175.

Hedges, R.E.M., Housley, R.A., Bronk Ramsey, C., Van Klinken, G.J., 1993. Radiocarbondates from the Oxford AMS system: archaeometry datelist 17. Archaeometry 35(2), 305326.

Hedges, R.E.M., Housley, R.A., Bronk Ramsey, C., Van Klinken, G.J., 1994. Radiocarbon

dates from the Oxford AMS system: archaeometry datelist 18. Archaeometry 36(2), 333374.

Hedges, R.E.M., Stevens, R.E., Koch, P., 2005. Isotopes and bones. In: Leng, M.J. (Ed.),Isotopes in Paleoenvironmental Research. Developments in Paleoenvir-onmental Research 10, 117145.

Hell, P., Paule, L., 1982. Ergebnisse taxonomischer Untersuchungen des Wolfes(Canis lupus) in den slowakischen Karpaten. Folia Zoological 31, 255270.

Hoefs, J., 1997. Stable Isotope Geochemistry. Springer, Berlin.Hofreiter, M., Serre , D., Poinar, H.N., Kuch, M., Paabo, S., 2001. Ancient DNA. Nature

Reviews Genetics 2, 353359.Katzenberg, A.M., Weber, A., 1999. Stable isotope ecology and Palaeodiet in

the Lake Baikal Region of Siberia. Journal of Archaeological Science 26,651659.

Kumar, S., Tamura, K., Nei, M., 2004. MEGA3: integrated software for molecularevolutionary genetics analysis and sequence alignment. Briefings in Bio-informatics 5, 150163.

Labunsky, A., 1994. The Central Asian hounds. The Travel Notes of a Cynologist.Dnepropetrovsk, Ukraine (in Russian).

Lawrence, B., 1967. Early domestic dogs. Zeitschrift fur Saugetierkunde 32,4459.



18/18

Lawrence, B., Bossert, W.H., 1967. Multiple character analysis ofCanis lupus, latrans,and familiaris, with a discussion of the relationships of Canis niger. AmericanZoologist 7, 223232.

Leonard, J.A., Vila, C., Fox-Dobbs, K., Koch, P.L., Wayne, R.K., Van Valkenburgh, B.,2007. Megafaunal extinctions and the disappearance of a specialized wolfecomorph. Current Biology 17, 11461150.

Leonard, J.A., Wayne, R.K., Wheeler, J., Valadez, R., Guillen, S., Vila, C., 2002. AncientDNA evidence for Old World origin of New World dogs. Science 298, 16131616.

Lindblad-Toh, K., Wade, C.M., Mikkelsen, T.S., Karlsson, E.K., Jaffe, D.B., Kamal, M.,Clamp, M., Chang, J.L., Kulbokas III, E.J., Zody, M.C., Mauceli, E., Xie, X., Breen, M.,

Wayne, R.K., Ostrander, E.A., Ponting, C.P., Galibert, F., Smith, D.R., deJong, P.J.,Kirkness, E., Alvarez, P., Biagi, T., Brockman, W., Butler, J., Chin, C.-W., Cook, A.,Cuff, J., Daly, M.J., DeCaprio, D., Gnerre, S., Grabherr, M., Kellis, M., Kleber, M.,Bardeleben, C., Goodstadt, L., Heger, A., Hitte, C., Kim, L., Koepfli, K.-P.,Parker, H.G., Pollinger, J.P., Searle, S.M.J., Sutter, N.B., Thomas, R., Webber, C.,2005. Genome sequence, comparative analysis and haplotype structure of thedomestic dog. Nature 438, 803819.

Miller, R., 2001. Lithic resource management during the Belgian Early UpperPalaeolithic: effects of variable raw material context on lithic economy. Etudeset Recherches Archeologiques de lUniversitede Liege 91.

Morey, D.F., 1986. Studies on Amerindian dogs: taxonomic analysis of canid craniafrom the Northern Plains. Journal of Archaeological Science 13, 119145.

Morey, D.F., 1992. Shape and development in the evolution of the domestic dog.Journal of Archaeological Science 19, 181204.

Nobis, G., 1986. Die Wildsaugetiere in der Umwelt des Menschen von Oberkassel beiBonn und das Domestikationsproblem von Wolfen im Jungpalaolithikum.Bonner Jahrbuch 186, 368376.

Olsen, S.J., 1985. Origins of the Domestic Dog: The Fossil Record. University ofArizona Press, Tucson.

Otte, M., 1979. Le paleolithique superieur ancien en Belgique. MonographiesdArcheologie Nationale 5.

Otte, M., Groenen, M., 2001. Le Paleolithique superieur en Belgique. Anthropologicaet Praehistorica 112, 3948.

Paabo, S., Poinar, H., Serre, D., Jaenicke-Despres, V., Hebler, J., Rohland, N., Kuch, M.,Krause, J., Vigilant, L., Hofreiter, M., 2004. Genetic analyses from ancient DNA.Annual Review of Genetics 38, 645679.

Parker, H.G., Kim, L.V., Sutter, N.B., Carlson, S., Lorentzen, T.D., Malek, T.B.,Johnson, G.S., DeFrance, H.B., Ostrander, E.A., Kruglyak, L., 2004. Geneticstructure of the purebred domestic dog. Science 304, 11601164.

Peterson, B.J., Fry, B., 1987. Stable isotopes in ecosystem studies. Annual Review ofEcology and Systematics 18, 293320.

Phillips, D.L., Koch, P.L., 2002. Incorporating concentration dependence in stableisotope mixing models. Oecologia 130, 114125.

Pidoplichko, I.G., 1998. Upper Palaeolithic dwellings of mammoth bones in theUkraine. In: BAR International Series 712.

Polikarpovich, K.M., 1968. Paleolit verhnego Podneprovya. Nauka i Technika, Minsk(in Russian).

Richards, M.P., Hedges, R.E.M., 1999. Stable isotope evidence for similarities in thetypes of marine foods used by late mesolithic humans at sites along the Atlanticcoast of Europe. Journal of Archaeological Science 26, 717722.

Richards, M.P., 2000. Human and faunal stable isotope analyses from Goats Holeand Foxhole Caves, Gower. In: Aldhouse-Green, S. (Ed.), Paviland Cave and theRed Lady A Definitive Report. University of Wales College and NationalMuseums and Galleries of Wales, Newport, pp. 7175.

Richards, M.P., Pettitt, P.B., Stiner, M.C., Trinkaus, E., 2001. Stable isotope evidencefor increasing dietary breadth in the European mid-Upper Paleolithic.Proceedings of the National Academy of Sciences of the United States ofAmerica 98, 65286532.

Richards, M.P., Hedges, R.E.M., 20 03. Variations in bone collagen d13C and d15Nvalues of fauna from Northwest Europe over the last 40,000 years. Palae-ogeography, Palaeoclimatology Palaeoecology 193, 261267.

Sablin, M.V., 2002. Paleozoologia stojanok verchnei decni: novie danie. In:Praslov, N.D. (Ed.), Verchnii paleolit verchnii pleistotsen: dinamika pripodnichsobimii i periodizatsija archeologiatsechix kultur. Russian Academy of Sciences,Institute of Material Culture, pp. 107109 (in Russian).

Sablin, M.V., Khlopachev, G.A., 2002. The earliest Ice Age dogs: evidence from Eli-

seevichi I. Current Anthropology 43, 795799.Sablin, M.V., Khlopachev, G.A., 20 03. Die altesten Hunde aus Eliseevici I (Russland).

Archaologisches Korrespondenzblatt 33, 309316.

Savolainen, P., Zhang, Y.-P., Luo, J., Lundeberg, J., Leitner, T., 2002. Genetic evidencefor an East Asian origin of domestic dogs. Science 298, 16101613.

Schoeninger, M.J., DeNiro, M.J., Tauber, H., 1983. Stable nitrogen isotope ratios ofbone collagen reflect marine and terrestrial components of prehistoric humandiet. Science 220, 13811383.

Schwarcz, H.P., Schoeninger, M.J., 1991. Stable isotopic analyses in human nutri-tional ecology. Yearbook of Physical Anthropology 34, 283321.

Shapiro, B., Drummond, A.J., Rambaut, A., Wilson, M.C., Matheus, P.E., Sher, A.V.,Pybus, O.G., Gilbert, M.T.P., Barnes, I., Binladen, J., Willerslev, E., Hansen, A.J.,Baryshnikov, G.F., Burns, J.A., Davydov, S., Driver, J.C., Froese, D.G.,

Harington, C.R., Keddie, G., Kosintsev, P., Kunz, M.L., Martin, L.D.,Stephenson, R.O., Storer, J., Tedford, R., Zimov, S., Cooper, A., 2004. Rise and fallof the Beringian steppe bison. Science 306, 15611565.

Shigehara, N., Hongo, H., 2000. Ancient remains of Jomon dogs from Neolithic sitesin Japan. In: Crockford, S.J. (Ed.), Dogs Through Time: an ArchaeologicalPerspective, Proceedings of the 1st ICAZ Symposium on the History of theDomestic Dog. BAR International Series 889, pp. 6167.

Soffer, O., 1985. The Upper Paleolithic of the Central Russian Plain. Academic Press,Orlando.

Stevens, R.E., Germonpre, M., Petrie, C.A., OConnell, T.C., Palaeoenvironmental andchronological investigations of the Magdalenian sites of Goyet Cave and Trou deChaleux (Belgium), via stable isotope and radiocarbon analyses of horse skeletalremains. Journal of Archaeological Science, in press, DOI:10.1016/j.jas.2008.10.008 .

Stevens, R.E., Hedges, R.E.M., 2004. Carbon and nitrogen stable isotope analysis ofNorthwest European horse bone and tooth collagen, 40,000 BP present:Palaeoclimatic interpretations. Quaternary Science Reviews 23, 977991.

Stevens, R.E., Jacobi, R., Street, M., Germonpre, M., Conard, N.J., Munzel, S.C.,Hedges, R.E.M., 2008. Nitrogen isotope analyses of reindeer ( Rangifer tarandus),45,000 BP to 9,000 BP: Palaeoenvironmental reconstructions. Palaeogeography,Palaeoclimatology, Palaeoecology, doi:10.1016/j.palaeo.2008.01.019.

Stiller, M., Green, R.E., Ronan, M., Simons, J.F., Du, L., He, W., Egholm, M.,Rothberg, J.M., Keats, S.G., Ovodov, N.D., Antipina, E.E., Baryshnikov, G.F.,Kuzmin, Y.V., Vasilevski, A.A., Wuenschell, G.E., Termini, J., Hofreiter, M., Jae-nicke-Despres, V., Paabo, S., 2006. Patterns of nucleotide misincorporationsduring enzymatic amplification and direct large-scale sequencing of ancientDNA. Proceedings of the National Academy of Sciences, USA 103, 13,57813584.

Street, M., 1989. Ein fruhmesolithischer Hund und Hundverbiss an Knochen vomFundplatz Bedburg-Konigshoven, Niederrhein. Archaologische Informationen12, 203215.

Tchernov, E., Valla, F.F., 1997. Two new dogs, and other Natufian dogs, from theSouthern Levant. Journal of Archaeological Science 24, 6595.

Tieszen, L.L., Fagre, T., 1993. Carbon isotopic variability in d13C modern andarchaeological maize. Journal of Anthropological Archaeology 20, 2540.

Ulrix-Closset, M., 1975. Le paleolithique moyen dans le bassin mosan en Belgique.Editions Universa, Wetteren.

Van Klinken, G.J., Van der Plicht, J., Hedges, R.E.M., 1994. Bone 13C/12C ratios reflectpalaeoclimatic variations. Geophysical Research Letters 21, 445448.

Van Neer, W., Wouters, W., Germonpre, M., 2007. Fish remains from three UpperPalaeolithic cave deposits in southern Belgium. Anthropologica et Praehistorica118, 522.

Van Valkenburgh, B., Wayne, R.K., 1994. Shape divergence associated with sizeconvergence in sympatric East African jackals. Ecology 75, 15671581.

Vigne, J.-D., 2005. Lhumerus de chien magdalenien de Erralla (Gipuzkoa, Espagne)et la domestication tardiglaciaire du loup en Europe. Munibe 57, 279287.

Vila, C., Amorim, I., Leonard, J., Posada, D., Castroviejo, J., Petrucci-Fonseca, F.,Crandall, K., Ellegren, H., Wayne, R., 1999. Mitochondri

Documents

Germonpre Et Al. Palaeolithic Dog-libre